libvirt-vmcfg 0.0.1

XML builder for libvirt virtual machine configuration

libvirt_vmcfg

libvirt_vmcfg is a library for generating XML VM specifications for use in libvirt.

It's in an alpha state but pull requests welcome.

Documentation is poor at the moment as I have other priorities. When I have time, I'll create better docs.

Examples

Domain builder

from libvirt_vmcfg.domain.profiles.linux_virtio import kvm_default_hardwar
from libvirt_vmcfg.domain.devices.interface import BridgedInterface
from libvirt_vmcfg.domain.devices.disk import (QemuDiskBlock, QemuDiskNet,
                                               DeviceType)
from libvirt_vmcfg.domain.util.disk import (TargetDevGenerator,
                                            qemu_driver_attrs_raw)
from libvirt_vmcfg.domain import Domain


# Automatic generator for target devices
t = TargetDevGenerator()

# Shorthand for creating the devices we need, based on what virt-install does
# This doesn't do everything - you still need disks and a network interface.
# You can also create your own hardware specification, see
# libvirt_vmcfg/profiles/linux_virtio.py for details.
elements = kvm_default_hardware(name="test", vcpus=2, memory=786432,
                                boot_dev_order=["hd"])

# This is the hard disk specification, using a block device
# Note how we use qemu_driver_attrs_raw here, these are the recommended
# parameters for a raw device with QEMU
hard_disk = QemuDiskBlock(device=DeviceType.DISK,
                          source_dev="/dev/foo-vg/lvm_disk_partition",
                          target_dev=t.next_virtio(),
                          driver_attrs=qemu_driver_attrs_raw)

# This is the CD specification
cdrom = QemuDiskNet(device=DeviceType.CDROM,
                    source_url="http://example.com/install/cloudinit-seed.iso",
                    target_dev=t.next_virtio(), readonly=True)

# Network interface, only bridging is supported right now
interface = BridgedInterface("br0")

# Obvious
elements.extend((hard_disk, cdrom, interface))

# Create our domain and emit the XML
# Since this is an example, we print it out
dom = Domain(elements=elements)
print(dom.emit_xml(pretty_print=True))

Volumes

from libvirt_vmcfg.volume import Volume


GIGABYTE = 1073741824

vol = Volume(name="test", size=3 * GIGABYTE)
print(vol.emit_xml(pretty_print=True))

TODO

A lot. I'll get around to most of these eventually. I hope. I'm sadly quite busy these days.

• More interface types
• More disk types, like LUNs
• More hardware profiles, like for Windows, old Linux, BSD, etc.
• More hardware that isn't virtio
• More hardware, period
• More hardware options, like more timers, etc.
• More hypervisors (Xen, VirtualBox, VMWare Workstation, and ESXi are probably the only other relevant ones)
  • Ensure hypervisor portability
• Investigate other architectures, like ppc64le and aarch64, add relevant bits from those
  • Ensure portability to said platforms
• Creation of XML for volumes
• Documentation
• Find a better way to handle driver_attrs with disks
• Find a clean way to add all the relevant doodads from the (very complicated) disk specification
• Make volume interoperate with domain where relevant

Contributions

See the contributing file for more information.

License

CC0, dedicated to the public domain. Do whatever you want with it.

Unfortunately I have to say this, because this comes up a lot in other projects. Let me make it clear: the project "licensing" (insofar a public domain dedication can be considered a license) is not up for debate. I wanted CC0 and it was a deliberate thoughtful choice. I am fully aware of its ramifications and have no intention to reverse my course on the matter. I was not — nor am not — willing or even able to enforce any restrictions on people or companies. Just do what you want with this. Relicensing is allowed; you may do so if you wish. Or don't. I don't care one way or the other.

Say thanks

You can thank me at my Say Thanks.

You can also donate to help keep me afloat during these tough times.